1、简介
MDC 全拼 Mapped Diagnostic Contexts，是SLF4J类日志系统中实现分布式多线程日志数据传递的重要工具（不同的系统有不同的实现方式，下文中会有介绍）；同时，用户也可利用MDC将一些运行时的上下文数据打印出来。
2、应用
自定义日志输出内容：用户可以将某个或某些所有日志中都需要打印的字符串设置于MDC中，这样就不需要每次打印日志时都专门写出来。例如：调用链系统中调用链唯一标示 traceID 及其中某次调用关系标示rpcID。
3、官方介绍
看一下logback中关于MDC的介绍：
Most real-world distributed systems need to deal with multiple clients simultaneously. In a typical multithreaded implementation of such a system, different threads will handle different clients. A possible but slightly discouraged approach to differentiate the logging output of one client from another consists of instantiating a new and separate logger for each client. This technique promotes the proliferation of loggers and may increase their management overhead.
大意为：现今分布式系统大多有多个客户端，使用多线程提供服务。为了在日志输出中体现出来自不同的客户端，一个轻量但不被鼓励的方式是为每个不同的客户端设置不同的日志。这种技术毫无疑问会极大的增加管理成本。
A lighter technique consists of uniquely stamping each log request servicing a given client. Neil Harrison described this method in the book Patterns for Logging Diagnostic Messages in Pattern Languages of Program Design 3, edited by R. Martin, D. Riehle, and F. Buschmann (Addison-Wesley, 1997). Logback leverages a variant of this technique included in the SLF4J API: Mapped Diagnostic Contexts (MDC).
大意为：另一个轻量级的实现方式是唯一的标示每个客户端发来的请求。Logback利用了SLF4J API中包含的一种技术：Mapped Diagnostic Contexts (MDC)。
看一下MDC类本身的注解：
If the underlying logging system offers MDC functionality, then SLF4J's MDC, i.e. this class, will delegate to the underlying system's MDC. Note that at this time, only two logging systems, namely log4j and logback, offer MDC functionality. For java.util.logging which does not support MDC, {@link BasicMDCAdapter} will be used. For other systems, i.e slf4j-simple and slf4j-nop, {@link NOPMDCAdapter} will be used.
大意为：目前只有log4j和logback提供原生的MDC支持，其他的会有其他的实现方式。
4、实现方式
此处以 Log4j2 中的实现为例。为了方便讲解，我们只分析MDC的put()方法：
public class MDC {
public static void put(String key, String val)
throws IllegalArgumentException {
if (key == null) {
throw new IllegalArgumentException("key parameter cannot be null");
}
if (mdcAdapter == null) {
throw new IllegalStateException("MDCAdapter cannot be null. See also "
+ NULL_MDCA_URL);
}
mdcAdapter.put(key, val);
}
MDC的put()方法利用MDCAdapter实现。
下面看一下Log4j中MDCAdapter的实现Log4jMDCAdapter：
public class Log4jMDCAdapter implements MDCAdapter {
public void put(String key, String val) {
ThreadContext.put(key, val);
}
}
public final class ThreadContext {
......
private static ThreadContextMap contextMap;
private ThreadContext() {
}
static void init() {
contextMap = null;
......
if(!useMap) {
contextMap = new NoOpThreadContextMap();
} else {
contextMap = ThreadContextMapFactory.createThreadContextMap();
}
}
public static void put(String key, String value) {
contextMap.put(key, value);
}
public static ThreadContextMap createThreadContextMap() {
// 此处获取用户自定义ThreadContextMap
final String threadContextMapName = managerProps.getStringProperty(THREAD_CONTEXT_KEY);
......(其他代码)
// 若无特殊设置，将使用默认ThreadContextMap
if (result == null) {
result = createDefaultThreadContextMap();
}
return result;
}
ThreadContextMapFactory 中 createDefaultThreadContextMap() 的定义：
private static ThreadContextMap createDefaultThreadContextMap() {
return (ThreadContextMap)(Constants.ENABLE_THREADLOCALS?(PropertiesUtil.getProperties().getBooleanProperty("log4j2.garbagefree.threadContextMap")?new GarbageFreeSortedArrayThreadContextMap():new CopyOnWriteSortedArrayThreadContextMap()):new DefaultThreadContextMap(true));
}
class GarbageFreeSortedArrayThreadContextMap implements ThreadContextMap2 {
public static final String INHERITABLE_MAP = "isThreadContextMapInheritable";
protected static final int DEFAULT_INITIAL_CAPACITY = 16;
protected static final String PROPERTY_NAME_INITIAL_CAPACITY = "log4j2.ThreadContext.initial.capacity";
//注意如下定义
protected final ThreadLocal<StringMap> localMap = this.createThreadLocalMap();
private ThreadLocal<StringMap> createThreadLocalMap() {
PropertiesUtil managerProps = PropertiesUtil.getProperties();
boolean inheritable = managerProps.getBooleanProperty("isThreadContextMapInheritable");
return (ThreadLocal)(inheritable?new InheritableThreadLocal<StringMap>() {
protected StringMap childValue(StringMap parentValue) {
return parentValue != null?GarbageFreeSortedArrayThreadContextMap.this.createStringMap(parentValue):null;
}
}:new ThreadLocal());
}
......
}
经过上述代码追踪可以看到，Log4j中定义了一个ThreadLocal，利用 ThreadLocal 实现不同请求的线程区分。如果对ThreadLocal不太了解，可以阅读ThreadLocal详解。
5、自定义ThreadContextMap
官方文档 中关于用户自定义ThreadContextMap的描述：
Any custom ThreadContextMap implementation can be installed by setting system property log4j2.threadContextMap to the fully qualified class name of the class implementing the ThreadContextMap interface. By also implementing the ReadOnlyThreadContextMap interface, your custom ThreadContextMap implementation will be accessible to applications via the ThreadContext::getThreadContextMap method.
让我们再看一下上文提到的createThreadContextMap()：
private static final String THREAD_CONTEXT_KEY = "log4j2.threadContextMap";
public static ThreadContextMap createThreadContextMap() {
final PropertiesUtil managerProps = PropertiesUtil.getProperties();
final String threadContextMapName = managerProps.getStringProperty(THREAD_CONTEXT_KEY);
final ClassLoader cl = ProviderUtil.findClassLoader();
ThreadContextMap result = null;
if (threadContextMapName != null) {
try {
final Class<?> clazz = cl.loadClass(threadContextMapName);
if (ThreadContextMap.class.isAssignableFrom(clazz)) {
result = (ThreadContextMap) clazz.newInstance();
}
} catch (final ClassNotFoundException cnfe) {
LOGGER.error("Unable to locate configured ThreadContextMap {}", threadContextMapName);
} catch (final Exception ex) {
LOGGER.error("Unable to create configured ThreadContextMap {}", threadContextMapName, ex);
}
}
......（其他代码）
}
createThreadContextMap()在开始处首先判断用户是否设置了系统变量log4j2.threadContextMap，若设置了，就开始实例化。
设置log4j2.threadContextMap变量有两种方式：
- 项目启动时，命令行中设置：-Dlog4j2.threadContextMap=xxxx
- 资源文件根目录下创建 log4j2.component.properties 文件，文件内填写 log4j2.threadContextMap=xxxx
6、总结
经过上述分析可以知道，Log4j中利用ThreadLocal实现MDC，达到不同线程间日志信息互不影响的目的。
